What are the advantages of using server grade processors

[cmaier]

That's over exaggerating the issue. You are making it seem like if you don't have ECC, your computer will go down every hour.

I can run my memory ~20% overclocked for long 100% loads at a time with F@H or Prime and I never get any issues.

From my experience, heat is the main killer of memory after proper installation. I'd rather take non-ECC ram that runs at a cool 25 C load than ECC at a presumable 55-65 C load.

I have never had any issues with non-ECC ram when maintained correctly, and IMO, you are making the issue of errors bigger than they should be.

Wow. You just aren't listening. I'm not talking about anything that has anything to do with how your RAM is "maintained." I am not talking about heat damage. I am not talking about anything that you can control. I'm not even talking about manufacturing defects or marginal defects. (Though all of these are factors).

The planet is constantly bombarded by alpha particles. (This is slightly different than cosmic radiation, but both are a problem). They're all around you. They come flying through space at near light speed and ram into stuff.

Your memory consists of billions of "RAM cells." Even if the chance of one particle hitting one RAM cell is tiny, there are so many RAM cells and so many particles that this happens all the time.

Under the right conditions, the energy from these collisions can excite electrons into the conduction band and can actually flip memory states. If you have one CPU to worry about, this probably happens fairly rarely. How often will depend on your altitude, among other factors. If you have a server room full of machines, this will happen much more often to at least one of them.

Now, when a bit flips, several outcomes are possible. First, it could be memory that is not in use. No harm done.

Second, it could be instruction memory. This may cause a crash (which would actually be GOOD news). This may cause the wrong instructions to get executed, but not cause a crash (this would be BAD news).

Third, it could be data memory. This will likely result in a calculation coming out wrong. Boom. Your bank account has negative dollars in it.

In mission-critical situations, you absolutely positively cannot accept the wrong answer. In even more mission-critical situations, you absolutely cannot even afford a crash (although, arguably, a consumer OS like mac os is not appropriate for that latter case).

ECC is critical exactly because these events are so rare. Since they happen infrequently (once a month, say, at sea level), ECC can correct them. (If they happened more frequently you would need a stronger code that could correct multi-bit errors).

I don't care how cool you keep your machine. I don't care if you swaddle it in baby blankets, sing it to bed at night, and buy it ice cream on its birthday. You aren't stopping those soft error events unless you've got ECC RAM.

 

[TheStrudel]

Good summary, cmaier. I think the stuff about bank accounts is particularly relevant in this day and age. It never really occurred to me before that omnipresent radiation could flip a bit that readily. Though my ECC memory probably has prevented a crash or three in video editing. Probably worth it with $$ at stake.

 

[jjahshik32]

Wow. You just aren't listening. I'm not talking about anything that has anything to do with how your RAM is "maintained." I am not talking about heat damage. I am not talking about anything that you can control. I'm not even talking about manufacturing defects or marginal defects. (Though all of these are factors).

The planet is constantly bombarded by alpha particles. (This is slightly different than cosmic radiation, but both are a problem). They're all around you. They come flying through space at near light speed and ram into stuff.

Your memory consists of billions of "RAM cells." Even if the chance of one particle hitting one RAM cell is tiny, there are so many RAM cells and so many particles that this happens all the time.

Under the right conditions, the energy from these collisions can excite electrons into the conduction band and can actually flip memory states. If you have one CPU to worry about, this probably happens fairly rarely. How often will depend on your altitude, among other factors. If you have a server room full of machines, this will happen much more often to at least one of them.

Now, when a bit flips, several outcomes are possible. First, it could be memory that is not in use. No harm done.

Second, it could be instruction memory. This may cause a crash (which would actually be GOOD news). This may cause the wrong instructions to get executed, but not cause a crash (this would be BAD news).

Third, it could be data memory. This will likely result in a calculation coming out wrong. Boom. Your bank account has negative dollars in it.

In mission-critical situations, you absolutely positively cannot accept the wrong answer. In even more mission-critical situations, you absolutely cannot even afford a crash (although, arguably, a consumer OS like mac os is not appropriate for that latter case).

ECC is critical exactly because these events are so rare. Since they happen infrequently (once a month, say, at sea level), ECC can correct them. (If they happened more frequently you would need a stronger code that could correct multi-bit errors).

I don't care how cool you keep your machine. I don't care if you swaddle it in baby blankets, sing it to bed at night, and buy it ice cream on its birthday. You aren't stopping those soft error events unless you've got ECC RAM.

Check and MATE.

 

[trancepriest]

Wow. You just aren't listening. I'm not talking about anything that has anything to do with how your RAM is "maintained." I am not talking about heat damage. I am not talking about anything that you can control. I'm not even talking about manufacturing defects or marginal defects. (Though all of these are factors).

The planet is constantly bombarded by alpha particles. (This is slightly different than cosmic radiation, but both are a problem). They're all around you. They come flying through space at near light speed and ram into stuff.

Your memory consists of billions of "RAM cells." Even if the chance of one particle hitting one RAM cell is tiny, there are so many RAM cells and so many particles that this happens all the time.

Under the right conditions, the energy from these collisions can excite electrons into the conduction band and can actually flip memory states. If you have one CPU to worry about, this probably happens fairly rarely. How often will depend on your altitude, among other factors. If you have a server room full of machines, this will happen much more often to at least one of them.

Now, when a bit flips, several outcomes are possible. First, it could be memory that is not in use. No harm done.

Second, it could be instruction memory. This may cause a crash (which would actually be GOOD news). This may cause the wrong instructions to get executed, but not cause a crash (this would be BAD news).

Third, it could be data memory. This will likely result in a calculation coming out wrong. Boom. Your bank account has negative dollars in it.

In mission-critical situations, you absolutely positively cannot accept the wrong answer. In even more mission-critical situations, you absolutely cannot even afford a crash (although, arguably, a consumer OS like mac os is not appropriate for that latter case).

ECC is critical exactly because these events are so rare. Since they happen infrequently (once a month, say, at sea level), ECC can correct them. (If they happened more frequently you would need a stronger code that could correct multi-bit errors).

I don't care how cool you keep your machine. I don't care if you swaddle it in baby blankets, sing it to bed at night, and buy it ice cream on its birthday. You aren't stopping those soft error events unless you've got ECC RAM.

I like this guy. Honor to have you on here with sentient input.

 

[notjustjay]

Third, it could be data memory. This will likely result in a calculation coming out wrong. Boom. Your bank account has negative dollars in it.

And that's the key difference between server class and consumer grade. If this happens to me on my laptop on a plane, I go "huh. that was weird" and reboot. Bottom line, I don't really care if it happens once in a blue moon.

If this happens in a mission-critical system, consequences can be grave. Damaging. Expensive. It must be prevented at all costs. Even (especially!) if it happens once in a blue moon, then you've got a "ghost in the machine" type of bug that's extremely difficult to isolate and fix, and if you're a vendor of mission critical systems, it is a Very Very Very Bad thing to have people describe your system as "Well, it works great 99.9% of the time, but every so often you do get that weird little glitch, nobody knows why or how to fix it."

"Oh, it happens so rarely that 99.99% of the time, you'll never see a problem" is just not good enough in this case.

 

[rylin]

Show me a Core i7 without SMP? Can you show me?

Tell me which 55xx series Xeon has more cache than the i7? Can you tell me?

Prove to me that i7 965EEs running at 90 degrees that have been stable for weeks at 100% load are somehow less stable than the Xeons? Can you prove it to me?


THAT is why all of what you said is stupid.

You do realize that this thread was a general question, not specific to the latest and greatest Intel have to offer, right? Right?

 

[kevink2]

Another reason is that Intel likes to place limitations on their "desktop" CPUs to that they don't compete too much with their server chips/motho use a motherboards. Support for multiple CPUs. Memory capacity.

In the not too distant past, if you wanted to use more than 4GB of physical RAM, you needed a higher end board. iMacs, for instance, were limited to 3 and 4 GB.

 

[nanofrog]

In the not too distant past, if you wanted to use more than 4GB of physical RAM, you needed a higher end board. iMacs, for instance, were limited to 3 and 4 GB.

I tend to see this as a choice of the board maker though, as they've a few chipsets to choose from (recent past). Now (Nehalem), the IMC is on the CPU. They didn't place any capacity limits between Core i7 and Xeon variants. Slot count can play a role, as does the specific memory type.

For example, some boards may be capable of running 96GB of Registered ECC (RDIMM's), while only 48GB of Unbuffered (UDIMM's) on 12 memory slots. The register chip on the RDIMM allows the capacity to double. Not the IMC.